Download UKCP09: Probabilistic projections of wind speed

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Instrumental temperature record wikipedia , lookup

Climate engineering wikipedia , lookup

Numerical weather prediction wikipedia , lookup

Media coverage of global warming wikipedia , lookup

Atmospheric model wikipedia , lookup

Politics of global warming wikipedia , lookup

Climate governance wikipedia , lookup

Citizens' Climate Lobby wikipedia , lookup

Economics of global warming wikipedia , lookup

Mitigation of global warming in Australia wikipedia , lookup

Climate change in Tuvalu wikipedia , lookup

Scientific opinion on climate change wikipedia , lookup

Public opinion on global warming wikipedia , lookup

Global warming wikipedia , lookup

Solar radiation management wikipedia , lookup

Climate change and agriculture wikipedia , lookup

Climate change and poverty wikipedia , lookup

Climate sensitivity wikipedia , lookup

Surveys of scientists' views on climate change wikipedia , lookup

Effects of global warming on humans wikipedia , lookup

Climate change feedback wikipedia , lookup

Attribution of recent climate change wikipedia , lookup

Years of Living Dangerously wikipedia , lookup

Climate change in the United States wikipedia , lookup

Climate change, industry and society wikipedia , lookup

General circulation model wikipedia , lookup

IPCC Fourth Assessment Report wikipedia , lookup

Transcript
UK CLIMATE
PROJECTIONS
UKCP09: Probabilistic projections of wind speed
UKCP09 additional product
http://ukclimateprojections.defra.gov.uk
2
UK CLIMATE
PROJECTIONS
Probabilistic projections of wind speed:
Executive summary
• Probabilistic projections of changes in 30-year mean wind speeds relative
to 1961–1990 have been produced for UKCP09, based on the same
methodology previously used to produce other variables available online.
As for other UKCP09 variables, the probabilistic climate projections of
changes in wind speed reflect current scientific understanding.
David M. H. Sexton and James
Murphy
Met Office Hadley Centre, Exeter
November 2010
• This report presents examples of the changes for the 2050s and 2080s.
The range of projected changes in summer wind speed for the 2050s
covers both positive and negative changes but is generally slightly skewed
towards negative changes. This is indicated by small reductions at most
UK locations in the 50% probability level, between 0 ms−1 and −0.2 ms−1.
A windspeed of 0.2 ms−1 is about 0.4 knots which is small compared with
the typical magnitude of summer mean wind speed of about 7–10 knots
over much of England and Northern Ireland, and 7–14 knots over much
of Scotland and Wales.
• In winter for the 2050s, the ranges of projected changes in wind speed are
approximately symmetric about near-zero change. The largest uncertainty
ranges are over Scotland, where the 10–90% probability range is typically
−0.5 ms−1 to 0.5 ms−1. This range equates to –1 to 1 knots, however these
changes in 30-year mean winter wind speeds are small compared with the
typical winter averages for present day climate, which are 10–14 knots
over the Lowlands and 18–24 knots over the Highlands.
• As for other variables in UKCP09, the projected changes are derived
principally from ensembles of perturbed variants of the Met Office
Hadley Centre (HadCM3) model. These show a variety of changes, due to
a spread of forced responses to increased greenhouse gas concentrations,
augmented considerably by natural climate variability. The forced
changes are influenced by several different factors, and vary between
different ensembles, and between individual ensemble members. While
the changes in large scale atmospheric circulation patterns over the
North Atlantic and Europe show some coherent features traceable to
basic aspects of the response to increased greenhouse gases, there is
little evidence of a consistent response in surface wind speed over the
UK. The spread due to uncertainties in the forced response is typically
larger in winter than in summer, but is always smaller than the spread
due to natural variability. These combined factors provide much of the
total uncertainty in the projections. Where the changes show a small shift
in the balance of probabilities towards lower or higher wind speeds, this
1
UKCP09: Probabilistic projections of wind speed
arises from the net impact of a variety of physical mechanisms captured
in the HadCM3 ensembles, and in a small ensemble of 12 international
climate models used to complement the sampling of model uncertainties
achieved via the HadCM3 simulations. However, these shifts are always
small in the context of the observed climatological mean wind speed.
• The results (as for all UKCP09 products) are subject to the caveat that
the current generation of climate models used to produce them could be
missing a key process liable to change the projections. In the case of wind
speed, some recent research suggests that including a better resolved
stratosphere in the climate model could alter the projected changes
over Europe, particularly in winter. However, the evidence should be
regarded as preliminary until a wider range of relevant modelling studies
are available. The next set of climate model projections being generated
around the world for the IPCC Fifth Assessment Report will include
several models containing improved representations of the effects of the
stratosphere on surface climate. Following provision and assessment of
these, it will be possible to assess the implications for the current results
in a more concrete fashion.
2
UK CLIMATE
PROJECTIONS
1 Background
The probabilistic projection methodology in UKCP09 involves sampling climate
modelling uncertainties by combining results from perturbed variants of the
HadCM3 configuration of the Met Office global climate model with projections
from an ensemble of alternative international climate models. However, as
discussed in the UKCP09 climate projections science report (Murphy et al. 2009), it
was not possible to provide probabilistic projections of future changes for certain
variables (soil moisture, latent heat flux, snowfall rate and wind speed). In the
case of wind speed, the required data was not available from the other global
climate models. This is because an accurate calculation of monthly mean wind
speed in a climate model has to be based on the accumulated values of wind
speed at individual model timesteps, and such diagnostics were not available for
the other global climate models. As an alternative, Brown et al. (2009) provided
advice on changes in wind speed based on the range of responses from an
ensemble of 11 variants of our own regional climate model that were used in the
production of UKCP09. Brown et al. (2009) concluded that the regional model
simulations did not sample the full range of possible outcomes, but that they
could be used to obtain a set of plausible future realisations of surface winds
(mean values and time series characteristics), once corrected for biases in the
simulated historical values.
However, after further investigation since the launch of UKCP09, we have
found that the required monthly mean wind speeds for the ensemble of other
climate models can be well approximated by accumulating values of wind speed
calculated from the daily means of the westerly and northerly components
of wind. This was demonstrated by estimating projected monthly wind speed
changes from time series of daily wind components for an ensemble of variants
of HadCM3, and comparing against the correct monthly wind speed changes
based on timestep-accumulated values. Figure 1 shows that the biases from this
approach are below 0.1 ms−1, which is small compared to the uncertainties in the
projected changes in section 3. Daily wind components are available for the 7 of
the 12 alternative climate models, allowing monthly wind speeds to be estimated
as described above. Monthly wind speed values for the remaining 5 models were
estimated from other variables, using inter-variable correlations obtained from
the ensemble of alternative models, following Murphy et al. (2009). 3
UKCP09: Probabilistic projections of wind speed
Figure 1: Ensemble mean bias (thick line)
in monthly mean wind speed calculated
from daily means of wind components for
17 configurations of HadCM3 (thin lines).
Bias in wind speeds derived from daily data
0.4
UK wind speed change (m/s)
0.3
0.2
0.1
0
–0.1
–0.2
J
F
M
A
M
J
Month
4
J
A
S
O
N
D
UK CLIMATE
PROJECTIONS
2 Large scale changes in circulation
Future changes in mean wind speed, and indeed in other climatological metrics
relating to storms or anticyclones (see Annex 6 of Murphy et al. 2009), can
potentially be influenced by several aspects of the forced response to increases
in greenhouse gases or aerosols. In particular, a number of factors likely to affect
future changes in mid-latitude cyclones have been identified in previous research
(e.g. McDonald, 2010). For example, increases in atmospheric moisture content in a
warmer atmosphere provide an increased source of energy for the intensification
of storms through condensation and precipitation, once the storms have formed.
In the upper troposphere, the warming is larger in the tropics than in polar
regions, tending to provide more energy for storm development. In winter, on
the other hand, the zonally averaged surface warming at high latitudes exceeds
that found at lower latitudes, tending to reduce the available energy. Also, the
land warms more than the oceans. In northern hemisphere winter, this reduces
the temperature contrasts which provide energy for the genesis of storms off
the eastern coasts of the continents. Regarding the UK and western Europe,
another factor is the relative minimum in the projected warming typically found
in central regions of the North Atlantic Ocean, particularly in model simulations
including a dynamic ocean component, since these generally predict a reduction
in the northward transport of warm water from the tropics associated with the
Atlantic Meridional Overturning Circulation (AMOC – see Annex 5 of Murphy
et al. 2009). In some (though not all) global model projections, this leads to a
southward shift in the storm track downstream of the consequent changes in the
north–south sea surface temperature gradient. The projected changes in wind
and storm characteristics over the UK in any particular model simulation will
therefore depend on a balance between several competing factors, and will in
addition be influenced by the unpredictable effects of natural climate variability.
For these reasons, the changes in circulation simulated in our ensembles of
perturbed variants of HadCM3 show considerable variability over the UK, which
are reflected in a wide spread of projected changes in mean surface wind speed.
In our largest ensemble, which simulates the equilibrium response to doubled
carbon dioxide concentrations using a simplified ocean, the ensemble-averaged
circulation changes in the troposphere (e.g. Figure 2a) show modest increases in
the mean westerly flow during winter in the Atlantic/European sector, between
about 40 and 60 degrees north. These increases are consistent with regional
increases in the north–south temperature gradient in the upper troposphere in
this latitude band, associated with the enhanced upper level warming simulated
in tropical regions (see above). However, the ensemble-averaged increases in
surface wind speed over the UK are small (not exceeding 0.2 ms−1). Furthermore,
there is a spread in the ensemble of projected changes: The forced component
5
UKCP09: Probabilistic projections of wind speed
Figure 2: Comparison of changes in
winter 850 hPa westerly wind speed
for a) ensemble mean of equilibrium
response to doubled CO2 of perturbed
variants of HadCM3, b) ensemble mean
of equilibrium response to doubled CO2
from other models, and c) ensemble mean
of transient response (2070–2099 minus
1961–1990) of variants of HadCM3.
a) Equilibrium response of 850 hPa westerly wind to doubled CO2 (HadCM3)
75N
60N
45N
30N
60W
–2
–1.5
30W
–1.0
–0.5
0
0
0.5
30E
1
1.5
2
b) Equilibrium response of 850 hPa westerly wind to doubled CO2 (other models)
75N
60N
45N
30N
60W
–2
–1.5
30W
–1.0
–0.5
0
0
0.5
30E
1
1.5
2
c) Transient response of 850 hPa westerly wind (HadCM3) 2070–2099 minus 1961–1990
75N
60N
45N
30N
60W
–2
6
–1.5
30W
–1.0
–0.5
0
0
0.5
30E
1
1.5
2
UKCP09: Probabilistic projections of wind speed
of the change is negative in some ensemble members, and uncertainties are
augmented considerably by the influence of natural climate variability, which
is larger than the spread of forced changes. The corresponding ensemble of
12 alternative climate models shows a similar large-scale pattern of changes in
atmospheric winds in its ensemble average (Figure 2b), however the band of
increased winds is displaced slightly to the north compared with the HadCM3
ensemble. This is probably because the enhancement in surface warming at high
latitudes, which tends to reduce the energy available for storm development, is
somewhat more pronounced in the HadCM3 ensemble. Over the UK, the multimodel ensemble therefore shows somewhat larger increases in lower tropospheric
winds over Scotland and northern England, resulting in slightly larger ensembleaveraged increases in the mean surface wind speed. Again, however, there is a
considerable spread in the responses of individual ensemble members.
In order to produce probabilistic projections for specific 21st century periods,
further simulations of time-dependent climate change are required, in addition to
the simulations of equilibrium climate change discussed above. In particular, the
time-dependent simulations included a dynamical ocean module, and therefore
account for the effects of ocean transport changes. In these simulations, also
consisting of an ensemble of variants of HadCM3 (see Murphy et al. 2009), a
smaller surface warming is predicted in the North Atlantic ocean relative to
the response over Greenland and the eastern seaboard of the north American
continent. This reduced warming is consistent with reductions in northward heat
transport via the AMOC, and reduces the energy for the genesis and development
of storms. This leads to an ensemble average response showing reductions in
the mean westerly wind over the Atlantic and western Europe, relative to the
equilibrium simulations (see Figure 2c). Therefore, over the UK, this mechanism
tends to offset the modest ensemble-averaged increases in surface wind speed
found in the simulations of equilibrium climate change. The probabilistic
projections reflect the combined sampling of these competing mechanisms of
change and their associated uncertainties, achieved by pooling the equilibrium
and time-dependent simulations. The results for winter therefore tend to show
distributions of changes centred on a near-zero response, as described below. In summer, ensemble average patterns of change in the simulations of equilibrium
climate change are consistent with a poleward shift in the summer storm track
found in previous climate model projections (Yin, 2005). This is associated with
the impacts of an enhanced equator-to-pole temperature gradient in the upper
troposphere. The patterns of changes in atmospheric westerly winds are broadly
similar between the HadCM3 and multi-model ensembles (Figure 3a cf 3b),
showing a band of increases extending eastwards from north-eastern Canada
across the Atlantic, and over most of the UK and northern Europe, with a band
of reduced westerlies to the south. Over the UK, this results in small increases
in the ensemble averaged changes in surface wind speed to the north, with
smaller increases (in the HadCM3 ensemble) or small decreases (in the multimodel ensemble) over southern England, which lies on the northern edge of
the band of reduced westerlies. In the ensemble of time-dependent HadCM3
projections a similar mean pattern of changes is found, however the northward
shift is slightly more pronounced than in the equilibrium simulations. Again this
is consistent with the effects of reductions in the AMOC, which limits the surface
warming in the north Atlantic, and hence tends to increase further the equatorto-pole contrast in the atmospheric response. The impact of ocean heat transport
changes therefore tends to offset the modest increases in westerly winds found
over most of the UK in the equilibrium simulations, since the zonal band of
increased westerlies is displaced to the north (Figure 3c). As in winter, the forced
7
UKCP09: Probabilistic projections of wind speed
Figure 3: Comparison of changes in
summer 850 hPa westerly wind speed
for a) ensemble mean of equilibrium
response to doubled CO2 of perturbed
variants of HadCM3, b) ensemble mean
of equilibrium response to doubled CO2
from other models, and c) ensemble mean
of transient response (2070–2099 minus
1961–1990) of variants of HadCM3.
a) Equilibrium response of 850 hPa westerly wind to doubled CO2 (HadCM3)
75N
60N
45N
30N
60W
–2
–1.5
30W
–1.0
–0.5
0
0
0.5
30E
1
1.5
2
b) Equilibrium response of 850 hPa westerly wind to doubled CO2 (other models)
75N
60N
45N
30N
60W
–2
–1.5
30W
–1.0
–0.5
0
0
0.5
30E
1
1.5
2
c) Transient response of 850 hPa westerly wind (HadCM3) 2070–2099 minus 1961–1990
75N
60N
45N
30N
60W
–2
8
–1.5
30W
–1.0
–0.5
0
0
0.5
30E
1
1.5
2
UKCP09: Probabilistic projections of wind speed
responses in surface wind speed corresponding to these changes in atmospheric
circulation are small. The forced responses also vary between ensemble members,
and natural variability again adds a considerable additional component of
uncertainty.
A basic assumption of our methodology for probabilistic projections is that our
largest ensemble of HadCM3 variants, used to simulate equilibrium climate
change as described above, provides a good first order estimate of the spread
of responses consistent with modelling uncertainties. This ensemble is then
augmented by use of the multi-model ensemble, in order to sample the effects of
plausible variations in model structure not considered in the HadCM3 ensemble
(see Chapter 3 of Murphy et al. 2009), and by further global and regional model
ensembles simulating time-dependent climate change. If the contribution
from the multi-model ensemble is too dominant (see Section 3.2.10 of Murphy
et al. 2009, then its potential lack of robustness (being based on a small set of
simulations assembled on an opportunity basis) could render the probabilistic
projections unreliable. It is important to check this for wind speed, as for variables
previously provided in UKCP09.
In general, the HadCM3 ensemble shows a wide spread of changes, with little
evidence of a systematic signal when averaged across the ensemble members.
Where the spread of responses does show a (small) systematic component, this
is found to be consistent with the larger scale patterns of circulation change (as
discussed above in the cases of summer and winter changes). In the multi-model
ensemble, the regional differences in circulation changes relative to the HadCM3
(see Figure 2 and related discussion) do alter the projected wind speed changes
over the UK to a limited degree. In general, the systematic component of the
wind speed changes across members of the multi-model ensemble tends to be
slightly larger than from the HadCM3 ensembles (though less so in summer), but
remains modest compared with the spread of outcomes implied by the HadCM3
ensemble. Therefore, we conclude that the multi-model ensemble plays its
expected role of modifying the information provided by the HadCM3 ensembles
to account for the effects of structural model uncertainties not sampled in the
latter. These impacts are modest, and are consistent with plausible uncertainties
in how different physical mechanisms combine to produce changes in wind speed
at the regional scale. We therefore conclude that production of probabilistic
projections based on the combined information from the HadCM3 and multimodel ensembles is justified. 9
UK CLIMATE
PROJECTIONS
3 Probabilistic projections
Like the other variables in UKCP09, the probabilistic projections of wind speed
were produced by combining results from the various model simulations outlined
above with a set of observational constraints to produce the projections, using
the statistical methodology described in Chapter 3 of Murphy et al. (2009).
Projections were produced for the same set of emissions scenarios, regions, times
of year and future periods described in section 1.2 of Murphy et al. (2009). Here
we provide a few illustrative examples.
Figure 4 shows that the range of anomalies in summer wind speed for the
2050s covers both positive and negative changes but is slightly skewed towards
negative changes, reflecting the combination of factors discussed earlier. The 50%
probability level shows a tendency towards a small reduction in wind speed, with
changes over most of the UK being between −0.2 ms−1 and 0 ms−1. A windspeed
of 0.2 ms−1 is about 0.4 knots which is small compared with the typical magnitude
of summer mean wind speed of about 7–10 knots over much of England and
Northern Ireland, and 7–14 knots over much of Scotland and Wales (Jenkins et
al. 2008). In the 2050s there is very little change in the projections with emission
scenario. By the 2080s (see Figure 5 ) the 10% probability level is more negative
than that of the 2050s with values most negative along the west coast of England
and Wales. For the 10% probability level, the minimum values are more negative
for the higher emissions scenario.
For winter in the 2050s (see Figure 6), the ranges of anomalies in wind speed
are quasi-symmetric about near-zero change. The 10–90% probability range is
typically around −0.5 ms−1 to 0.5 ms−1 with a small variation in magnitude across
the different emission scenarios. The spread in changes in wind speed is greatest
over Scotland, with somewhat larger ranges of uncertainty of −0.7 ms−1 to
0.75 ms−1 are seen in the high emission scenario compared with the low emission
scenario (about −0.4 ms−1 to 0.5 ms−1). This range is roughly –0.8 to 1 knot and is
relatively small compared with observed values of 10–14 knots over the lowlands
of Scotland and 18–24 over much of the Highlands. We note that the spread of
possible changes shown by the probability distributions is strongly influenced
by natural climate variability. However, the spread is somewhat wider than that
implied by natural variability alone (this is also true of the changes in other
seasons), despite the lack of significant shifts in the values of the 50% probability
level. This is partly due to the simulation of a spread of plausible forced responses
in the ensembles of model projections, as discussed in Section 2. Uncertainties
arising from the statistical techniques used to convert the model simulations into
probabilistic projections also contribute to the spread of outcomes. 10
UKCP09: Probabilistic projections of wind speed
50% probability
Central estimate
Figure 4: 10, 50, and 90% probability level
maps of changes in summer wind speed
(ms−1) for the 2050s (2040–2069) relative
to 1961–1990 for the Low (bottom row),
Medium (middle row), and High (top row)
emission scenarios.
90% probability
Very unlikely to be
greater than
Low emissions
Medium emissions
High emissions
10% probability
Very unlikely to be
less than
–1.4
–1.2
–1
–0.8 –0.6 –0.4 –0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Change in windspeed (ms )
–1
11
UKCP09: Probabilistic projections of wind speed
50% probability
Central estimate
Figure 5: 10, 50, and 90% probability level
maps of changes in summer wind speed
(ms−1) for the 2080s (2070–2099) relative
to 1961–1990 for the Low (bottom row),
Medium (middle row), and High (top row)
emission scenarios.
90% probability
Very unlikely to be
greater than
Low emissions
Medium emissions
High emissions
10% probability
Very unlikely to be
less than
–1.4
–1.2
–1
–0.8 –0.6 –0.4 –0.2
Change in windspeed (ms )
–1
12
0
0.2
0.4
0.6
0.8
1
1.2
1.4
UKCP09: Probabilistic projections of wind speed
In the 2080s (Figure 7) the 50% probability level shows small reductions in wind
speed of about −0.1 ms−1 over the UK, except for the Highlands where small
increases are found. The results again reflect the range of competing mechanisms
and their associated uncertainties discussed earlier (see Jenkins et al. 2008).
We note that the changes in 30 year averages of wind speed presented here are
only one of several possible metrics relating to future regional characteristics
of the atmospheric circulation and its variability. Annex 6 of Murphy et al.
(2009) reports changes in a number of diagnostics relating more specifically
to variability associated with storms and anticyclones over the UK, though the
general conclusion remains that it is difficult to identify consistent signals of
increase or decrease.
All probabilistic climate projections in UKCP09 reflect the current understanding
of the climate system as represented in the current state-of-the-art models, and
it is possible that processes missing from all these climate models might affect
the projections. For wind speed, this is particularly pertinent, as a recent study
(Huebener et al., 2007) suggests that the forced response of wind speeds over
Europe could be significantly affected by how well the stratosphere is resolved
in the climate models. Scaife et al. (2005) demonstrates that the simulated
characteristics of low frequency variability associated with the North Atlantic
Oscillation could also be improved. However, further research is needed
to confirm the effects of troposphere–stratosphere interactions on future
projections of European climate. The next set of projections being produced for
the IPCC Fifth Assessment Report will include a number of climate models with a
better resolved stratosphere. Once these become available, it will be possible to
make a firmer assessment. 13
UKCP09: Probabilistic projections of wind speed
50% probability
Central estimate
90% probability
Very unlikely to be
greater than
Low emissions
Medium emissions
High emissions
10% probability
Very unlikely to be
less than
–0.8 –0.7 –0.6 –0.5 –0.4 –0.3 –0.2 –0.1
Change in windspeed (ms )
–1
14
0
0.1
0.2 0.3
0.4 0.5 0.6
0.7 0.8
Figure 6: 10, 50, and 90% probability level
maps of changes in winter wind speed
(ms−1) for the 2050s (2040–2069) relative
to 1961–1990 for the Low (bottom row),
Medium (middle row), and High (top row)
emission scenarios.
UKCP09: Probabilistic projections of wind speed
50% probability
Central estimate
90% probability
Very unlikely to be
greater than
Figure 7: 10, 50, and 90% probability level
maps of changes in winter wind speed
(ms−1) for the 2080s (2070–2099) relative
to 1961–1990 for the Low (bottom row),
Medium (middle row), and High (top row)
emission scenarios.
Low emissions
Medium emissions
High emissions
10% probability
Very unlikely to be
less than
–0.8 –0.7 –0.6 –0.5 –0.4 –0.3 –0.2 –0.1
0
0.1
0.2 0.3
0.4 0.5 0.6
0.7 0.8
Change in windspeed (ms )
–1
15
UKCP09: Probabilistic projections of wind speed
Acknowledgements
References
We thank Mat Collins and Hazel
Thornton for extracting the daily
data from the PCMDI archive, and
for validating the use of this data
to approximate monthly mean
windspeeds.
Brown, S., Boorman, P., McDonald, R.
& Murphy, J. 2009. Interpretation and
use of surface wind speed projections
from the 11-member met office
regional climate model ensemble.
Technical report, Met Office Hadley
Centre.
Huebener, H., Cubasch,U., Langematz,
U., Spangehl, T., Niehorster, F., Fast, I.
& Kunxe, M. 2007. Ensemble climate
simulations using a fully coupled
ocean-troposphere-stratosphere
general circulation model.
Philosophical Transactions of the
Royal Society London, 365, 2089–2102.
Jenkins, G. J., Perry, M. C. & Prior, M.
J. 2008. The climate of the UK and
recent trends. Met Office, Hadley
Centre, Exeter, UK.
McDonald, R. E. 2010. Understanding
the impact of climate change
on northern hemisphere extratropical cyclones. Climate Dynamics,
Doi:10.1007/s00382-010-0916-x.
16
Murphy, J. M., Sexton, D. M. H.,
Jenkins, G. J., Booth, B. B. B., Brown,
C. C., Clark, R. T., Collins, M., Harris, G.
R., Kendon, E. J., Betts, R. A., Brown,
S. J., Humphrey, K. A., McCarthy,
M. P., McDonald, R. E., Stephens,
A., Wallace, C., Warren, R., Wilby,
R. & Wood, R. A. 2009. UK climate
projections science report: Climate
change projections. Met Office Hadley
Centre, Exeter, UK.
Scaife, A. A., Knight, J. R., Vallis, G. K.
& Folland, C. K. 2005. A stratospheric
influence on the winter NAO and
North Atlantic surface climate.
Geophysical Research Letters, 32,
L18715. Doi:10.1029/2005GL023226.
Yin, J. H. 2005 .A consistent poleward
shift of the storm tracks in simulations
of 21st century climate. Geophysical
Research Letters, 32, L14701.
Doi:10.1029/2005GL023684. 22.